What is the difference between "per deveice" and "per transistor" thermal resistance? We can see for example in the datasheet of BC846BPN: www.nxp.com/documents/data_sheet/BC846BPN.pdf (page 4, section 6).

For example, if the power dissipation of one of the transistors is P1, and that of the other is P2, then what is the temperature rise:

DT_a=R_TH_perDevice*(P1+P2)

or

DT_b=R_TH_perTransistor*P1+R_TH_perTransistor*P2 = R_TH_perTransistor*(P1+P2) ???

Unfortunately I can not understand why these two parameters are defined, and in which case which one should be used?

Thanks for the answers in advance.

If you use one transistor, then you should use "Rth per transistor" if you use two transistors with approx equal power dissipation then you should use "Rth per device". If power dissipations of the transisors are different, then you can estimate temperature rise using following formula:

DT = Rth * SQRT (P1^2 + P2^2);

where Rth is "per transistor", P1 and P2 power dissipations on transistors. This is not an exact formula, however gives good approach (about 6% accuracy in the worst case).

Regards,

Bulat